My AI Pro subscription has pretty much become completely unusable since Gemini 4. I get 3 to 5 basic prompts before I hit my quota on Gemini 4 with any level of thinking. Switching back to 3.8 Flash which is what I had previously been able to use all day and only hit 2-5 percent of my quota now quickly exceeds my quota in just 10 to 15 prompts.
October 15, 2026
Today I hit the limits in a large chat with 550k–600k tokens:
-
Gemini 3.1 Pro – after about 20–30 generations
-
Gemini 3.8 Flash – after about 20–30 generations
Naturally, the models continue to work normally in new and small chats.
-
If this is a deliberate change, I don’t understand why the models aren’t hitting the limit everywhere, but only in large chats. Furthermore, I don’t understand why the system assumes that processing tokens on Flash models costs about the same as on Pro models. In theory, token processing on Flash models is cheaper than on Pro models, which means the limit should be higher—and that’s how it used to be before September: there were about 50 generations on Pro models and 100+ generations on Flash models, as I recall.
-
The theory is still out there that this is a bug causing the quota system to stumble in the cache of long chats
I don’t know what else to say. Google is silent; it hasn’t confirmed whether this behavior is intended or not. I’m starting to get tired of this.
Please, Pro and Ultra subscribers: if you’re running into a limit in long chats, post the approximate limit number, the length of the context, the model, and your subscription type.
Yes, Argon in the Gemini App uses up a fairly large percentage of the quota.
It seems like a bug because right now when I check my usage for the 5 hour window I am at 8% and weekly I am at 29% but I’m completely locked out from doing anything because every chat window I have been working in pops up a window saying I’m over my quota and can’t continue until my quota resets or I upgrade.
Perhaps there has been a misunderstanding, but my complaints (and not just mine, but those of all users here) are about the limits in AI Studio, where there isn’t even a 5-hour limit, let alone the ability to view limit usage in Playground.
After Gemini 4 Argon was released, I left AI Studio and now use Gemini 4 Argon exclusively through the Gemini app. The Gemini app isn’t bad either, so I recommend giving it a try. At the very least, you can see how much of your quota you’ve used and how it’s being consumed.
It also doesn’t have some of AI Studio’s limitations, such as the lack of support for certain file types and languages. It fully supports things like CUDA and TypeScript. On top of that, Argon can connect to additional tools and services, such as Google Drive and the workspace environment, which I think makes it better than AI Studio.
To sum it up, I’ve migrated to the Gemini app because I found it to be the more practical choice. And honestly, given the access it provides to Argon, I’d recommend it to you as well.
I don’t know, but based on my personal observations, the models in Google AI Studio perform much better than those in the Gemini App. Perhaps this is because AI Studio doesn’t have as many predefined instructions and restrictions as the Gemini App does.
No misunderstanding. The limitations happen across products using the models. It’s just easier to monitor and document usage going through the Gemini Chat/Tasks interface.
Yes, the fact that the Gemini App lets you track limits is a huge advantage over Playground in AI Studio. Still, the limit systems in Gemini and AI Studio are different.
Today’s results for Gemini 3.1 Pro in large chats.
-
1 successful generation in a chat for 150k tokens
-
5 successful generations in a chat for 940k tokens
-
Around 10 successful generations in a chat for 600 tokens, after which I hit the limit in that chat
This is extremely frustrating
A quick update. In the chat with 150,000 tokens, 3.1 Pro generated another response. I don’t know how many more it will generate.
But Argon seems to have the same problem as Gemini 3.5–3.8 Flash.
Chat metrics:
-
Input tokens: 183,515
-
Output tokens: 762,598
Model performance:
-
Gemini 2.5 Pro (before it was disabled) — generates a response
-
Gemini 3.1 Pro—generates a response
-
Gemini 3.5 Flash (before it was disabled), 3.6 Flash, 3.7 Flash (before it was disabled), 3.8 Flash, and 4 Argon—return an error stating there are too many input tokens
I am genuinely wondering what they have done. Because I am also noticing with my Ultra sub that Gemini 4 Argon has a variable use limit based on the tokens in chat.
I also go that same limit regarding token usage rather than actual prompts. If it has been changed from strait prompts per day to a maximum token use per day this is a massive issue.
Mainly because they have clearly and massively changed the value of the subscription. And because they have done it with no reasonable notice. I’m quite sure we got the notice of change like the day of implementation. Google is strait up in breach of contract/consumer protection law. ToS does NOT protect in this case at all. We could sign anything within the ToS, but if it breaks actual contract/consumer protection laws they are non enforceable.
@Google Team
You are required to provide reasonable notice for any material or value changes in a payed service. Subscriptions are a payed service. Your x2, x5 and x20 is already manipulative and misleading because you have no actual baseline to provide users. You have given no valid metrics for users to look at for these multiplications.
October 18, 2026
A chat with 450k tokens. Gemini 3.0 Flash was only able to generate about 20 messages before hitting the limit.
Even if the new limit system simply counts tokens, it doesn’t seem to do so consistently.
Unacceptable usage limits on paid PRO/Advanced account
I am a paying Gemini Advanced / Pro subscriber, and I am writing to file a formal complaint regarding severe and unacceptable usage limits. I am currently locked out and forced to wait several hours after sending just a few prompts.
This is a complete breach of trust. To make matters worse, the rate limit was triggered because the AI kept failing to deliver proper answers, forcing me to waste prompts on systemic errors. I am not on the free tier; I am paying for a premium service, and being hit with hours of downtime makes it completely unusable.
I
If I want to talk about my status today with the AI Pro subscription with the Gemini 4 Argon model in AI studio today I had close to about 10 20 requests with Argon in a chat of about 400k tokens and well I have not reached the daily usage limit yet and well it is a good situation for me how about you guys?
is that Gemini 4 Argon available for Pro Users in AI Studio ? I still didn’t see it yet. It asks for Ultra subscription to use in AI Studio
Why it still shows unlock Gemini 4 Argon in the left interface ?